CVE-2026-34755: vLLM Denial of Service via Unbounded Base64 Frames

# CVE-2026-34755: vLLM Denial of Service via Unbounded Base64 Frames

https://gcdradar.vercel.app/incidents/cve-2026-34755-vllm-denial-of-service-via-unbounded-base64-f · AIR-2026-0555 · closed 2026-04-03

- Lane: Patch now
- Severity: 2 (Disclosed vulnerability)
- Fix status: Fix status unknown
- First reported: 2026-04-03
- AI role: Attack on AI
- Products: vLLM
- CVE ids: CVE-2026-34755
- CVSS: 6.5

## Summary

An out-of-memory vulnerability in vLLM allows attackers to crash the server via an unbounded base64-encoded video frame input. The vendor was notified and the issue is documented in a security advisory.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2026-04-06: CVE-2026-34755: vLLM is an inference and serving engine for large language models (LLMs). (https://nvd.nist.gov/vuln/detail/CVE-2026-34755)
- 2026-04-03: CVE-2026-34755: vLLM: Denial of Service via Unbounded Frame Count in video/jpeg Base64 Processing (https://github.com/advisories/GHSA-pq5c-rjhq-qp7p)

## Sources

- 2026-04-03 GitHub Advisory Database: CVE-2026-34755: vLLM: Denial of Service via Unbounded Frame Count in video/jpeg Base64 Processing https://github.com/advisories/GHSA-pq5c-rjhq-qp7p
- 2026-04-06 NVD: CVE-2026-34755: vLLM is an inference and serving engine for large language models (LLMs). https://nvd.nist.gov/vuln/detail/CVE-2026-34755

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.